[sdiy] The annual I/O impedance/protection thread! (revised)

[Neil Johnson]

Hi,

>>Outputs
>> http://www.sdiy.org/juz/euro_output_03.pdf
>
>>And for the non-inverting circuits add anti-parallel
>>Schottky diodes across the op-amp inputs and a resistor to the +ve input.
>
> I've added Schottky diodes based on advice from you and Harry B - it's also
> been suggested that I could use 1N4148s.
> Can you clarify what you mean about the resistor to the +ve input? Where
> to/from? Why? Value based on?

In the non-inverting case, definitely on the path to the +ve input to
the op-amp before the diodes to limit any fault current that might
flow from the signal source either into the op-amp or through the
diodes.  Better still a matching resistor on the connection to the -ve
pin.  10k would be enough.

> You'd ditch the zeners and their associated protection resistor from the
> output as well?

Well, since you seem unable to provide any *specific* requirements for
what you are hoping to protect the outputs from, I'm assuming that the
worst case is an input of -15V and the op-amp trying to push +15V.
All you need to do then is put some resistance, say 1k0, in the path
to limit the fault current.

>>Summing
>> http://www.sdiy.org/juz/euro_summing_03.pdf
>
>>2/ R1 will appreciably load the input pot to give a non-linear control
>>curve.  This might not be an issue for you (designers choice) but it
>>irks me.
>
> So how would you lay out an attenuated input for an inverting summing config
> or what would you suggest revising about that input with regards to
> component choice, value, order ,etc?

Its something I'm working on:

http://www.milton.arachsys.com/nj71/index.php?menu=2&submenu=2&subsubmenu=4

>>5/ Depending on what the internal load looks like that 1N4001 could
>>behave in a rather non-linear fashion.  And assuming enough current
>>passes through it to turn it on, it will drop about 1V from the output
>>of the op-amp.
>
> Yes, agreed, but considering it will probably be feeding something along the
> lines of my beloved single transistor VCA - it does the job just fine. When
> I need a precision solution I'll use a 'proper' rectifier.

Well, that's your call.  And its the first time you've given any
information about what these circuits are supposed to be driving.

>>6/ When the 5V1 zener turns on its going to provide a nice low-impedance
>>path to ground to pull lots of current out of the op-amp, which will
>>overheat and possibly melt.  Not quite desirable.
>
> I've used this config plenty of times and not even noticed any undue warmth
> - and I do check. Certainly never had one melt.

What was the op-amp output pin sitting at when you did the overvoltage test?

> Your impedance point does concern me though. Considering that the aim of the
> 1N4001/5V1 Zener combo is to allow an AC or negative V input to be used as a
> CV source, then attenuated, biased, blocked and capped so you only get e.g.
> 0-5V - what would you suggest for capping the upper DC limit?

A zener in the op-amp feedback path would be a better starting point.

Neil
-- 
http://www.njohnson.co.uk